Aluminum Probe Pad Thickness and Properties for Stable Parametric Probe Ability

Transcription

1 Aluminum Probe Pad Thickness and Properties for Stable Parametric Probe Ability Presenter: Bernd Bischoff (TI Germany) Co authors: Jerry Broz (ITS) Swen Harder (FFI) Uwe Schiessl (TI Germany) Seyedsoheil Khavandi (ITS) Gernot Zieschang (HTT)

2 Background Experienced poor life time of Takumi probe cards High amount of Al debris on the probe tips High amount of cleaning touchdown online Additional cleaning offline / outside Unstable CRES performance Different behavior seen for different pad metal thickness

3 Overview Objectives Perform Benchtop Testing using TI Supplied Wafers Execute TDs on Al wf to assess accumulation and CRES stability for different Al thickness Execute TDs on Al wf + Cleaning to assess cleaning efficiency Apply insight and learning from Benchtop testing to Production Environment Materials Takumi dutlet with two wired pins was provided by HTT/FFI Dresden Four wire CRES testing was performed across two pins Aluminum wafers supplied by TI Freising FAB representative of technology nodes Al thickness = 6kA / 15kA / 30kA Methods Apply overdrives matched to production On wafer: electrical measurements performed from first touch to (1) 30um and (2) 60um On cleaning material: cleaning performed at 60um Assess cleaning performance for CRES and debris removal

4 Controlled Test Conditions ITS Bench top System for material characterization and probe performance testing NI LabVIEW Motion Control Data Acquisition Overview Variable z speed and z acceleration. Synchronized load vs. overtravel vs. CRES data acquisition. High resolution video imaging and still image capture. Thermal chuck for elevated temperature testing. Current forcing and measurement with Keithley 2400 source meter. ITS LTU Probe Gen System

5 Benchtop CRES Measurement 2 wires to probe Sacrificial Probe (not used for CRES) 2 wires to probe 4 wire Kelvin CRES Measurement

6 Test Setup Overview DUTLET Aluminum Wafer Stepping Pattern FFI Takumi Test Vehicle Current = 10mA Application = 10msec Temperature = 25C

7 Imaging of Takumi TDs

8 6kÅ/15kÅ/30kÅ Thick Aluminum Wafer 6kÅ thick Aluminum Layer Small grain structures with some hillock like features EDS shows aluminum and titanium species 15kÅ thick Aluminum Layer Large grain structures EDS shows aluminum and trace titanium species 30kÅ thick Aluminum Layer Very large grain structures EDS shows aluminum and no trace titanium species

9 Benchtop Methods for Quantifying CRES Improvement CRES vs. Touchdown Charts Scatter plots demonstrate unstable CRES after multiple touchdowns Advantages: Demonstrate CRES stability during wafer test Indicative of when cleaning is required to reduce CRES Disadvantages: Difficult to assess incremental changes in CRES behavior Cumulative Frequency Distribution (CFD) Charts CFD plots show the observations falling in (or below) a specified limit and the shape of the curve indicates CRES stability level Advantages: Provides an easy way to compare different large data sets Incremental changes in CRES behavior can be identified Width of CRES distribution can be determined Disadvantages: Does not include a time component

10 CRES Performance Charts 6kA Al Wafer with Overtravel = 30um Pin to Pin Contact Resistance (ohms) Pin to Pin Contact Resistance (ohms) 7.50 Unstable CRES OT = 30um Aluminum Wafer TDs 6kA Al Wafer with Overtravel = 60um 7.50 Stable CRES OT = 60um Cummulative Frequency (%) CRES Stability Improvement Stable CRES Unstable CRES N = Contact Resistance (ohms) Aluminum Wafer TDs

11 6kÅ Thickness Wafer No Cleaning 120 CRES Stability Improvement N = 2500 Wfr Overdrive = 30um 100 Cummulative Frequency (%) um OT 30um OT Scrub Direction Wfr Overdrive = 60um Contact Resistance (ohms)

12 15kÅ Thickness Wafer No Cleaning Wfr Overdrive = 30um 120 N = Cummulative Frequency (%) um OT 30um OT Scrub Direction Wfr Overdrive = 60um Contact Resistance (ohms)

13 30kÅ Thickness Wafer No Cleaning Wfr Overdrive = 30um 120 N = Cummulative Frequency (%) um OT 30um OT Scrub Direction Wfr Overdrive = 60um Contact Resistance (ohms)

14 Probe Mark Assessment 6kÅ Wafer 15kÅ Wafer 30kÅ Wafer OT = 30um Length = ~10 to ~14um Depth = ~3.5 to ~3.8kÅ OT = 30um Length = ~12 to ~14um Depth = ~8.0 to ~8.3kÅ OT = 30um Length = ~16 to ~18um Depth = ~14.6 to 16.2kÅ OT = 60um Length = ~18 to ~20um Depth = ~5.0 to 5.4kÅ OT = 60um Length = ~20 to ~22um Depth = ~12.0 to 12.3kÅ OT = 60um Length = ~22 to ~24um Depth = ~20.0 to 22.3kÅ

15 First summary Three wafers received TI Freising 6kA, 15kA, and 30kA thick aluminum SEM showed differences in the grain sizes, structures, and boundaries. 6kA had small grains and a large number of grain boundaries 15kA had large grain structures and fewer grain boundaries 30kA has very large grain structures CRES performance and debris accumulation was different between wafers 6kA showed unstable CRES with a large amount of debris build up. 15kA showed relatively stable CRES with some debris build up. 30kA showed relatively stable CRES with minor debris build up. Probe marks size and depth differed.

16 Objectives Cleaning Matrix Overview Assess cleaning performance vs. Metal thickness Obtain cleaning effects / efficiency with minimum cleaning parameter Implement Probe Polish 150 (PP150) to maximize cleaning efficiency Cleaning recipes executed for 30um and 60um probing OD PP 150 / 30um production OD 6kA = 100TD interval with PP150 at 60um with 5TD per cycle. 15kA = 1000TD interval with PP150 at 60um with 5TD per cycle. 30kA = 1000TD interval with PP150 at 60um with 5TD per cycle. cleaning TDs 6kA Aluminum Wafer with Cleaning OD = 60um Cleaning Interval x3 times repeated 50TDs 100TDs 200TDs 500TDs 1000TDs 5 TDs x 1 x 10 TDs 20 TDs reference x 40 TDs PP 150 / 30um production OD 15kA Aluminum Wafer with Cleaning OD = 60um 30kA Aluminum Wafer with Cleaning OD = 60um cleaning TDs Cleaning Interval x3 times repeated 50TDs 100TDs 200TDs 500TDs 1000TDs 5 TDs x 1 10 TDs 20 TDs reference x 40 TDs PP 150 / 60um production OD cleaning TDs Cleaning Interval x3 times repeated 50TDs 100TDs 200TDs 500TDs 1000TDs 5 TDs x 1 x 10 TDs 20 TDs reference x 40 TDs PP 150 / 60um production OD cleaning TDs Cleaning Interval x3 times repeated 50TDs 100TDs 200TDs 500TDs 1000TDs 5 TDs x 1 10 TDs 20 TDs reference x 40 TDs

17 kÅ Thickness Wafer PP150 Cleaning Performed Wfr Overdrive = 30um CRES Stability Improvement N = 2500 Cleaning = 60um (100 / 5) Cummulative Frequency (%) um OT (no clean) 30um OT (100TD / 5TD with PP150) 60um OT (no clean) 60um OT (100TD / 5TD with PP150) Contact Resistance (ohms) Scrub Direction Wfr Overdrive = 60um Cleaning = 60um (100 / 5)

18 kÅ Thickness Wafer PP150 Cleaning Performed CRES Stability Improvement N = 5000 Wfr Overdrive = 30um Cleaning = 60um (1000 / 5) Cummulative Frequency (%) um OT (no clean) 30um OT (1000TD / 5TD with PP150) 60um OT (no clean) 60um OT (1000TD / 5TD with PP150) Contact Resistance (ohms) Scrub Direction Wfr Overdrive = 60um Cleaning = 60um (1000 / 5)

19 kÅ Thickness Wafer PP150 Cleaning Performed CRES Stability Improvement N = 5000 Wfr Overdrive = 30um Cleaning = 60um (1000 / 5) Cummulative Frequency (%) um OT (no clean) 30um OT (1000TD / 5TD with PP150) 60um OT (no clean) 60um OT (1000TD / 5TD with PP150) Contact Resistance (ohms) Scrub Direction Wfr Overdrive = 60um Cleaning = 60um (1000 / 5)

20 Conclusions / Recommendations I Conclusion Pad aluminum properties (6kA, 15kA, and 30kA) affected CRES stability and debris generation for two different production level overdrives (OT = 30um and 60um) Pad aluminum thickness and properties will have direct affect on the overall probe process Pad aluminum thickness dependent cleaning regimes are needed for optimal performance and to meet production requirements as well as maximize the probe card lifetime Cleaning recipes Improved CRES stability at the lower overtravel values Collected / controlled debris from the pyramid and contact area Small amounts of aluminum residuals on the contact surfaces

21 Conclusions / Recommendations II CRES results and probe tip surface inspection (based on aluminum wafers studied in the project) 5 clean TDs are insufficient and should be increased 100TD interval for 6kA is sufficient for debris control 1000TD interval for 15kA for improved debris collection; more frequent execution suggested 1000TD interval for 30kA might be sufficient for controlling debris Recommendations to increase probe card lifetime from 55TD interval / 40 clean TD recipe 20TDs per cleaning cycle + 100TD interval for 6kA ~4X potential probe card lifetime extension over 55TD interval 20 TD per cleaning cycle + 500TD interval for 15kA ~5X potential probe card lifetime extension over 55TD interval 20TDs per cleaning cycle + 500TD interval for 30kA ~5X potential probe card lifetime extension over 55TD interval

22 Production test with new cleaning conditions

23 CRES data vs. metal thickness and cleaning conditions

24 Tip life TD Head 5: ~10um Head 8: ~ 8 10um Head 7: ~ 7 10um Head 2: ~ 15 19um TD TD TD TD Head 6: ~11um Head 8: ~ 10 11um Head 7: ~ 9um Head 2: ~ 16 20um TD TD TD TD

25 Expected life TD increase from geometry Expected life TD limited by tip hight ~ min. 2.5Mio Expected life TD limited by tip diameter ~ min. 3.5Mio

26 Expected life TD increase from number of cleaning TDs

27 Expected cost reduction Cost factor = 1/ (pin# * TD#)

28 Summary / Conclusion From bench top experiments on Alu wafer Pad Alu thickness / test overdrive dependent CRES behavior Cleaning recipes for improved CRES stability and debris control Expectation to increase lifetime by min. 4X From production test with recommended cleaning recipes More stable CRES behavior Dramatically less number of online cleaning TD Less maintenance work, increase maintenance interval by 2X Till now less damage on pyramids Life time expectations is ~ min. 2.5X to 3.5X Cost reduction expectation is ~ 2.5x to 3.5x

29 Follow On Work Finalize production life time test Stable CRES Life TD vs. Cleaning TD Tip geometry EOL expected cost reduction achieved? Apply pad aluminum thickness dependent cleaning regime to other probe card technologies for parametric and wafer probe

30 Thanks to: Acknowledgement ITS: Jerry Broz & Seyedsoheil Khavandi FFI: Swen Harder HTT: Gernot Zieschang TI: Uwe Schiessl & process engineers